Lecture 26: Strong Anticipation and Direct Perception (Turvey, 2019, Lectures on Perception)

In this final chapter, Turvey lays out the basics of the ecological approach to coordinating action with respect to the future. This is actually a key part of behaviour - we reach for things we know are graspable before grasping them, we move to intercept or avoid things before they are anywhere near us, and so on. The question is how do we do this.

The ecological approach cannot, of course, use prediction to solve this problem. Prediction entails representation, making educated guesses about what is coming up based on loans of intelligence we cannot ever pay off. The whole point of this book has been to lay out the argument that a) this approach is doomed to fail but that b) the ecological approach, grounded in laws, is different enough in kind to be a viable option. 

Coordinating with things not in the immediate present is therefore going to rely on lawful informational coupling to dynamical events, with several interesting consequences. 

The Cartesian worldview we are rejecting comes with four relevant fallacies that constrain how those theories can interact with the future. 

1. Subjective time: time is not invented by the organism. Instead, time is defined by the dynamics of events, and because what is happening now in an event is lawfully linked to what just happened and what will happen, the event can provide information about all of this. 
2. Sensation-based perception:  the future is a problem for these accounts because the future is not currently happening, and so there's no light from it being sent to the eye. If, however, perception is based in information, then there can be currently available information about things that have happened, things that are happening, and things that will happen (within the bounds of the dynamics identifying the event). 
3. Perception assigned to the present: the traditional approach based in sensations thinks that perception must therefore be about what is happening now, because that's what creates the sensations. Information makes it possible for 'the present' to be more than a knife-edge slice of time, and to span the duration of the event.
4. Anticipation assigned to the future: if perception is contained to the knife-edge present, then anticipation requires representation of the currently absent future. If perception is based in information of events, however, anticipation can be something based on what is currently available in the perceptual array. 

The ecological approach, that perception is based in information of event dynamics, allows for strong anticipation: anticipation not via simulation or models, but via the laws of the dynamical event in question. 

To make this concrete, Turvey discusses several key empirical examples. He describes progressive occlusion (one surface coming in or out of view behind another as you or the surfaces move) and how the information in the occluding edge specifies the occlusion, the continued existence of the disappearing surface, and the future existence of the appearing surface. The outfielder problem and its various informational solutions is another key example, as well as Lee's general tau theory. 

Each of these demonstrate that the causal structure of the underlying event (e.g. one surface moving out from behind another as you approach) creates information that specifies that event structure forward and backwards in time, up to the limits of the underlying event dynamics. Strong anticipation becomes a corollary of direct perception based on information - they are deeply related, and possibly simply identical proposals. 

Turvey ends with a quote from Gibson:

Explanations of perception based on sensory inputs fail because they all come down to this: In order to perceive the world, one must already have ideas about it. Knowledge of the world is explained by assuming that knowledge of the world exists. Whether the ideas are learned or innate makes no difference; the fallacy lies in the circular reasoning.

But if, on the other hand, perception of the environment is not based on a sequence of snapshots but on invariant–extraction from a flux, one does not need to have ideas about the environment in order to perceive it.

(Gibson, 1979/1986, p. 304)

This is the Lectures in a nutshell: any theory that leads to a loan of intelligence has failed. Only a theory based in laws can explain perception.